Article pubs.acs.org/JPCB
Intergrowth and Interfacial Structure of Biomimetic Fluorapatite− Gelatin Nanocomposite: A Solid-State NMR Study Anastasia Vyalikh,*,† Paul Simon,‡ Elena Rosseeva,‡,§ Jana Buder,‡ Rüdiger Kniep,‡ and Ulrich Scheler*,† †
Leibniz-Institut für Polymerforschung Dresden e.V., Hohe Strasse 6, 01069 Dresden, Germany Max-Planck-Institut für Chemische Physik fester Stoffe, Nöthnitzer Strasse 40, 01187 Dresden, Germany § University of Konstanz, Physical Chemistry, POB 714, D-78457 Konstanz, Germany ‡
S Supporting Information *
ABSTRACT: The model system fluorapatite−gelatin allows mimicking the formation conditions on a lower level of complexity compared to natural dental and bone tissues. Here, we report on solid-state NMR investigations to examine the structure of fluorapatite−gelatin nanocomposites on a molecular level with particular focus on organic−inorganic interactions. Using 31P, 19F, and 1H MAS NMR and heteronuclear correlations, we found the nanocomposite to consist of crystalline apatite-like regions (fluorapatite and hydroxyfluorapatite) in close contact with a more dissolved (amorphous) layer containing first motifs of the apatite crystal structure as well as the organic component. A scheme of the intergrowth region in the fluorapatite−gelatin nanocomposite, where mineral domains interact with organic matrix, is presented.
■
INTRODUCTION The biomimetic system consisting of fluorapatite (FAp) and gelatin closely resembles the biosystem hydroxyapatite− collagen, which is intimately associated with calcified hard tissues in the human body, such as bone and teeth.1,2 However, using the model system of FAp−gelatin permits us to mimic the formation conditions on a lower level of complexity than the natural formation of dental and bone tissues, and therefore contributes to a deeper understanding of the principles of biomineralization. Substitution of the rigid, insoluble collagen by water-soluble gelatin (denatured collagen) holds the chance of easy rearrangement of the protein fibers during the formation of the organic−inorganic nanocomposite. Calcium phosphate composites containing gelatin and other proteins have been extensively investigated during recent years.3−11 Fluorapatite−gelatin composite aggregates are grown by double diffusion of buffered calcium and phosphate fluoride solutions in a 300 bloom gel of gelatin. Details for their preparation are described elsewhere.12,13 Characterization using different experimental methods has shown that the resulting material is composed of hierarchically ordered spherical aggregates with fractal architecture and diameters ranging from 140 to 200 μm. The chemical composition determined by elemental analysis and FTIR spectroscopy corresponds to Ca5‑x/2(PO4)3‑x(HPO4)x(F1‑y(OH)y), 2.3 wt % gelatin, x = 0.82, 0 < y ≤ 0.1. Synchrotron X-ray diffraction on individual aggregates in different states of growth has provided information on the orientation of the mineral component.14 Local variations in the degree of mineralization across a sample have been investigated by transmission electron microscopy.15 © 2013 American Chemical Society
Particular attention has been paid to the study of pattern formation as a function of time (morphogenesis). It has been found that aggregate formation starts with elongated hexagonal prismatic seeds, followed by fractal branching and development of growing dumbbell states, and finally ends up with a slightly notched sphere.16 The “controlling factors” for the fractal development that induces the formation of dumbell-like composite structures are mainly given by the intrinsic electrical fields as proven by electron holography.7,17 So far, investigations have been extensively focused on microscopy techniques. However, a detailed study of how the inorganic phase of fluorapatite−gelatin nanocomposite interacts with the organic phase at a molecular level is lacking. The present work is a contribution to the molecular-level characterization of fluorapatite−gelatin nanocomposites in order to get molecular information on structure formation of biomimetic organic−mineral nanocomposites. Solid-state nuclear magnetic resonance (NMR) has been applied, which is sensitive to the local environment of a particular nucleus, and, therefore, is suited to study interfacial phenomena on a molecular level, whereas other structural techniques characterize the ″long-range″ order, thus giving only an average view of the structure. In this study 31P, 19F, and 1H MAS NMR and their heteronuclear combination have been performed to examine the structure of the fluorapatite−gelatin nanocomposites. Received: October 17, 2013 Revised: December 17, 2013 Published: December 19, 2013 724
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
■
Article
of 500.1 MHz for 1H, 470 MHz for 19F, and 202.5 MHz for 31P. P MAS NMR spectra were acquired with high-power proton TPPM decoupling18 using a 3.2 mm MAS probehead. Spinning frequencies of 2 kHz, 5 kHz, and 10 kHz, a single 90°-pulse of 3.2 μs duration, a recycle delay of 80 s, and 64 repetitions were applied. For the 31P{1H} cross-polarization (CP) measurements contact times of 100 to 1500 μs and a recycle delay of 5 s were used. The 19F ultrafast MAS NMR experiments were acquired using a 1.3 mm probehead with a spinning frequency of 60 kHz, a 90°-pulse duration of 5 μs, and a recycle delay of 40 s. 1 H and 19F MAS NMR spectra at 30 kHz spinning speed were measured using a 2.5 mm HFX-MAS probehead. The 19F MAS NMR experiments were acquired with a 90°-pulse duration of 5 μs and a recycle delay of 40 s. For 1H MAS NMR, the 90°-pulse duration of 3 μs, a recycle delay of 5 s, and high-power 19F decoupling using XiX composite pulse19 and rf field of 80 kHz were applied. The two-dimensional 19F−1H HETCOR experiment was performed at 30 kHz MAS using frequency-switched Lee−Goldburg (FSLG) cross-polarization with contact times of 0.1 and 1.5 ms. A recycle delay of 5 s and 1024 scans per t1 time increment were used. A total of 64 t1 slices with 22.75 μs increment in the indirect dimension were acquired. Prior to Fourier transform, apodization with 120-Hz and 30-Hz exponential line broadening was used in 19F and 1H dimensions, respectively. 1H chemical shifts were referenced to tetramethylsilane (TMS) at 0 ppm using poly(vinylidene fluoride) as an external reference; powdered ammonium dihydrogen phosphate was used to reference the 31P spectra at 0.72 ppm relative to 85 wt % H3PO4. The 19F chemical shifts were referenced relative to CFCl3 at 0 ppm using PTFE as an external reference (δF = −122 ppm). The spectra were fitted using Dmfit.20
EXPERIMENTAL SECTION Synthesis of the Samples. In a typical double diffusion experiment,6,14,15 pig skin gelatin (300 bloom, Aldrich) was heated together with water (gelatin concentration maximum 15%) under stirring for a few minutes in a water bath (after acidification with 2 N HCl). The central tube of the diffusion cell was filled with hot gelatin and left at room temperature in a vertical position until gelation was finished (for ca. 24 h). The L-shaped tubes were then attached to the central gel-tube, filled with aqueous solutions (e.g., 0.133 m CaCl2 and 0.08 m Na2HPO4/0.0027 m KF), and closed with ground glass stoppers. Both solutions were previously adjusted to the physiological pH of 7.4 with α,α,α-tris(hydroxymethyl) methylamine/HCl. The diffusion experiments were carried out between 25 and 30 °C (thermostat). Chemical Analysis. The composite aggregates were ground, washed 3 times for 20 min in distilled water at 40 °C, then centrifuged and finally dried at 40 °C in order to remove the fraction of gelatin, which is only physisorbed on the aggregates surface and not integrated in the composite. The content of calcium, phosphate, and sodium was determined by inductively coupled plasma-optical emission spectrometry (ICPOES, Varian, VISTA RL). Ten milligram samples were dissolved in 4 mL of 2 M HCl and then diluted with water in a 100 mL volumetric flask. The fluoride content was determined by using F− ion selective electrode (Mettler Toledo Inc., Wilmington, MA). The amounts of carbonate and gelatin were calculated on the basis of the C and N contents determined by use of the carrier gas hot extraction method combined with the combustion technique by means of the CHNS-analyzer 932 (LECO, USA). The sample weighting amounted to 2 mg and combustion was performed in Sn capsules under O 2 atmosphere. X-ray Diffraction. X-ray powder data were collected in transmission mode using a Huber G670 Image Plate Camera, Cu Kα1 radiation (λ = 1.540 598 Å) and germanium (111) monochromator. Lattice constants a and c of apatite were calculated by least-squares refinements using LaB6 (cubic, a = 4.156 92 Å) as internal standard. FT-IR Spectroscopy. Fourier transform infrared (FT-IR) spectra in the region of 4000−400 cm−1 were recorded at room temperature using a Bruker spectrometer (IFS 66v/S; Globar (MIR), KBr, DTGS-Detector; Program Opus/IR 3.0.3). The samples were prepared as KBr pellets (1 mg of the material under investigation dispersed in 150 mg KBr). In order to separate the overlapping bands the spectral ranges were fitted with Pseudo-Voigt 1 distributions for the bands. The software package Origin 7.0 (OriginLab Corporation, MA, USA) was used for the calculations. Raman Spectroscopy. For Raman measurements ground powder was spread on a glass plate. Raman spectra in the region of 1500−300 cm−1 were collected at room temperature using a LabRam System 010 (Horiba Jobin Yvon, France) in backscattering mode. The 632.817 nm line of a He−Ne laser with 1 mW light power was used for excitation of samples under a microscope (20-fold magnification). The spectrometer was calibrated using Si as a reference. In order to separate the overlapping bands the spectral ranges were fitted with PseudoVoigt 1 distributions for the bands. Spectra were fitted in Origin 7.0 (OriginLab Corporation, MA, USA). NMR. All NMR spectra were obtained on a (11.7 T) Bruker Avance III 500 spectrometer operating at resonance frequencies
31
■
RESULTS P MAS and 31P {1H} CP MAS NMR. Figure 1a shows the 31 P MAS NMR spectrum of FAp-gelatin nanocomposite displaying a single line centered at 2.3 ppm. Based on the literature data21−23 this signal is attributed to PO43− groups in crystalline fluorapatite. Its line width of 1.1 ppm is typical for that of pure HAp and deviates significantly from that we observed in hydroxyapatite-based nanocomposite prepared by direct precipitation in solution11 (2.3 ppm). The broadening of the latter was explained by the presence of an amorphous calcium phosphate phase located at the surfaces or at grain boundaries of the HAp nanoplatelets. To investigate the structures, which are formed at the mineral−organic interface, we applied cross-polarization (CP) NMR, which is based on the magnetization transfer from 1H to 31P sites and enables emphasizing the 31P species with strong dipole−dipole coupling to protons. The 31P{1H} CP MAS NMR spectrum measured at a spinning speed of 2 kHz to retain the information on the chemical shift anisotropy is shown in Figure 2. Fit data for three components and their assignment are summarized in Table 1. The spectrum is dominated by a peak at 2.5 ppm attributed to PO43− groups in crystalline fluorapatite. Its small chemical shift anisotropy (CSA) of about 16 ppm is a result of a high symmetry in an isolated PO4 tetrahedron reflected by four nearly identical P−O bond lengths. Two weaker shoulder peaks at 1.3 ppm and 4.9 ppm have been assigned to protonated HPO42− and unprotonated PO43− phosphate surface groups, respectively, according to the previous assignment in synthetic 725
31
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
calcium phosphates, supported by the difference in their CSA values.21,24 The smaller CSA value of −17 ppm for unprotonated PO43− (surface) reflects the most likely tetrahedral phosphorus coordination. This orthophosphate group can be located at the mineral surface. In contrast, the larger CSA for the protonated HPO4 sites (62 ppm) points to the distortion of symmetry compared to PO43− tetrahedra yielding either HPO4 (surface) or HPO42− (bulk) groups, when the latter may replace the PO43− (bulk) in the apatite crystal structure. The CP build-up curves for each spectral component resolved in the CP spectra (Figure 1c) confirm the proposed assignment demonstrating the slower build-up for the peak at 2.5 ppm characteristic of apatite PO43− groups in contrast to other two signals, which reach the maximum at ca. 1 ms showing closer proximity to the protons. Complementary studies have proven the existence of HPO4 groups, which appeared as small peaks and shoulders around 874 cm−1 and 1006 cm−1 in the Raman spectrum (see SI Figure S1) and as peaks around 545 cm−1 and 864 cm−1 in the IR spectrum (see SI Figure S2). Chemical analysis of the FAp−gelatin nanocomposite has revealed the amount of hydrogen phosphate in the bulk phase to be about 10 mol %, which is not in agreement with 62% found from NMR (see Table 1). NMR indeed reflects the amount of phosphorus-containing species in the near-surface or interface region, but not the composition of the entire sample. This becomes visible if one compares the centerband 31P CP NMR spectrum at 10 kHz (Figure 1b) with the single pulse (SP) spectrum (Figure 1a). The asymmetric lineshape and significant broadening of the 31P CP spectrum has been explained by the selectivity of the CP experiment, which detects only a fraction of the 31P nuclei, which is in close proximity to protons, and whose fraction is significantly smaller with respect to all phosphorus atoms in the bulk detected in the 31 P SP experiment. Therefore, upon proper scaling of intensity the SP spectra shall completely conceal the tiny contribution from 31P atoms, which are observed in the CP spectra. 19 F and 1H MAS NMR. The 19F MAS spectrum at 60 kHz is shown in Figure 3a. Upon deconvolution, four components are discerned, whose fit parameters are summarized in Table 2. The peak at −102.8 ppm is known to arise from crystalline fluorapatite.23 The second component at −103.8 ppm can be assigned to apatite-like structures containing hydroxyl groups or isolated water molecules in apatite channels, based on the correlation of 19F chemical shift parameters and the F−/OH− ratio in mixed hydroxyfluorapatite reported in ref 25. The similarity of the shift tensor parameters of two peaks discussed above points to their similarities in the crystal structure and the local 19F environment. In contrast, the strong increase of the 19 F line width of the latter (4.9 ppm vs 1.6 ppm in pure crystalline fluorapatite) results from an inhomogeneous broadening due to structural disorder caused by the presence of the proton-containing species. The broad peak at −108.5 ppm is typical for Ca···F interactions, for example, in CaF2,26 although
Figure 1. 31P MAS (10 kHz) NMR spectra of FAp−gelatin nanocomposite detected using (a) direct polarization and (b) crosspolarization from 1H together with the deconvolution results. Spectra are normalized to a maximum intensity. The CP spectrum is fitted with three spectral components with the same isotropic shift parameters as used in the slow MAS spectrum (Table 1, Figure 2), but with the varied intensities and linewidths caused by folding back the spinning side bands; (c) CP build-up for three components indicated in Table 1 and (b) (squares denote the signal at 1.3 ppm, circles at 2.5 ppm, triangles at ca. 5 ppm).
Figure 2. Broad width 31P{1H} CP MAS NMR spectrum at a spinning speed 2 kHz and 1 ms contact time of FAp−gelatin nanocomposite (top) and its deconvolution (bottom). The green line denotes the component assigned to apatitic PO43−; the red and blue ones arise from unprotonated and protonated surface phosphate groups, respectively. The difference between the generated fit and the experimental spectrum is shown underneath.
Table 1. Fit Data of 31P{1H} CP MAS NMR Spectruma chemical shift tensor, ppm
a
δiso
δani
η
width, ppm
I, %
assignment
region
1.3 ± 0.3 2.5 ± 0.1 4.9 ± 0.3
62 ± 5 16 ± 3 −17 ± 3
0.4 ± 0.2 0.5 ± 0.2 0.2 ± 0.1
4.1 ± 0.2 2.3 ± 0.2 2.6 ± 0.5
62 ± 5 27 ± 5 11 ± 5
HPO42− PO43− PO43−
bulk/surface bulk surface
MAS 2 kHz, contact time 1 ms. 726
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
Figure 4. 19F−1H HETCOR spectrum of FAp−gelatin nanocomposite. The spectra on the top are 19F MAS and the horizontal projection of the 2D spectrum. On the right are the single-pulse 1H MAS (30 kHz) spectrum and the 1H projection of the 2D spectrum.
Table 3. 1H Isotropic Chemical Shifts δH (±0.1 ppm) as Measured from 1H−19F HETCOR with Correlation to the 19 F Chemical Shifts (δH, ppm /δF, ppm) and 1H MAS NMR
Figure 3. (a) 19F MAS (60 kHz) NMR spectrum of FAp-gelatin nanocomposite. (b) Projection (total signal intensity) of the 19F−1H HETCOR spectra in 19F direction at contact time of 0.1 and 0.5 ms. Colored lines represent decomposition and the gray line denotes the difference between the generated fit and the experimental spectra.
1 H−19F HETCOR
2.9/−103.1 6.6/−103.6 6.6/−110.5 7.1/−85.5 7.1/−95.5 11.1/−99 ca. 13/−110
Table 2. Fit Parameters of 19F MAS NMR Spectrum at 60 kHz chemical shift tensor, ppm δiso (±0.1)
δani (±5 ppm)
−95.7 −102.8 −103.8 −108.6
35 52 53 35
η
width, ppm (±0.5)
intensity, % (±5)
assignment
0.5 0.7 0.7 0.5
9.2 1.6 4.9 5.1
6 18 49 27
organic phase fluorapatite hydroxyfluorapatite amorphous phase
1
H MAS NMR
Assignment
1.1 and 1.5 2.5 5.7 6.1 6.5−10.0 -
organic phase or isolated water isolated water close to apatite-like phase water bound to organic phase water bound to organic phase fluorinated organic phase fluorinated organic phase strongly H-bonded protons HPO4 (surface/bulk) groups
containing (OH, HPO4) fluorapatite structures. In contrast, the peak at −102.8 ppm, which is observed in the 19F SP MAS NMR spectrum (Figure 3a), does not appear here, confirming its assignment to pure crystalline FAp. In the 2D spectrum, it is easy to see that both 19F signals at −103.6 ppm and −110.5 ppm correlate with the 1H peak at 6.6 ppm. The chemical shift of the 1H peak is too high to be attributed to bulk or surface water, but could be related to water molecules strongly hydrogen bonded to, most probably, protein molecules. Additionally, two other overlapping signals, however, with well-defined maxima arise at 7.1 ppm/−85.5 ppm and 7.1 ppm/−95.5 ppm. For their assignment, we refer to literature data, where close 1H signals in relevant systems have been observed. Thus, in 1H CRAMPS spectra of bovine collagen and animal bone27 as well as in a 1H−31P HETCOR spectrum of synthetic hydroxyapatite−gelatin nanocomposite reported in our previous work,11 the peaks at ca. 7 ppm have been assigned to the organic phase. Therefore, the correlation peaks at −85.5 ppm and −95.5 ppm detected in the present work can arise either from the fluorinated gelatin or from new structures on the mineral surface caused by the presence of the organic phase. Finally, the minor peak observed at 2.9 ppm/−103.1 ppm is assigned to isolated water molecules in the crystalline FAp, i.e., apatite-like channel water. It should be mentioned that similar 1 H signals at 3.2 ppm and 6.6 ppm have also been reported for low−OH− fluor−chlorapatite28 and were attributed to 1H atoms in a separate phase, not associated with apatite. Comparison of the HETCOR projections in the 19F direction at two different contact times (Figure 3b) yields the information about 1H− 19 F dipole interactions and an
no indication for the presence of CaF2 has been found in the corresponding Raman spectrum. Additionally, X-ray powder diffraction data are also consistent with the absence of CaF2 (see SI Figure S3). The line broadening of the signal at −108.5 ppm may originate from structural disorder resulting in a broad distribution of isotropic chemical shifts, as well as from strong homo- and heteronuclear dipolar coupling. Since application of high power 1H decoupling and the variation of the MAS speed (see SI Figure S4), expected to evidence the presence of the dipolar interactions, did not exhibit an effect on the linewidth, we concluded that structural disorder is the main origin of the inhomogeneous line broadening. To investigate the nature of the peaks at −108.5 ppm and −95.7 ppm, a further twodimensional 19F−1H heteronuclear correlation (HETCOR) experiment has been performed, which provides information on the spatial correlation of 1H containing groups in the inorganic phase and organic matrix to the 19F nuclei initially assumed to be present exclusively in the mineral phase. The two-dimensional 19 F− 1 H HETCOR spectrum (Figure 4) shows one asymmetric broad cross signal with the maximum intensity at δH/δF = 6.6 ppm/−110.5 ppm. For accurate analysis of the 2D spectrum we inspected individual slices in both directions. The chemical shifts of the correlation peaks are summarized in Table 3. The second dominating contribution is visible at 6.6 ppm/−103.6 ppm. We have to emphasize that the appearance of the peak at −103.6 ppm in the 2D experiment confirms its assignment to hydrogen 727
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
gelatin exposed to fluoride solution (see SI Figure S5a) is similar to that of protonated gelatin,34 whereas the 1H{19F} CP NMR experiment (see SI Figure S5b) enables resolving the 1H peaks at 7.2 and 11.6 ppm, which are very close to those observed in the HETCOR data (7.1 and 11.1 ppm). This observation corroborates their assignment to the organic phase. Fluorination of gelatin fibers has been proven in the 19F spectra (see SI Figure S5c,d). Although a very poor signal-to-noise ratio resulting from a low degree of fluorination hinders unequivocal assignment of 19F resonances in gelatin, several 19F signals can be distinguished. Comparing the SP and CP 19F spectra (Figure S5c,d), one can see that the signal at ca. −98 ppm shows the closest proximity to protons and thus can be attributed to the organic phase. This result confirms our assignment of the HETCOR data presented in Table 3 and supports a scenario of partial fluorination of gelatin during the formation of the nanocomposite.
estimation of associated protons-to-fluorine proximities, and therefore provides a cross-check toward the assignment of 19F signals. Apparently, the signals at ca. −85 ppm, −95 ppm, and −110 ppm have faster cross-polarization build-up translating the stronger 1H−19F dipole interactions than that in the apatitelike phase (at ca. −103 ppm). Thus, estimation of the build-up of the former signals points to 1H···19F distances shorter than 2 Å. It is worth to note that the distances of spatially separated F− and OH− ions in substituted fluorapatite are 2.03 Å for a F··· F···H group and 2.12 Å to 2.18 Å for a H···F···H groups along the crystallographic c axis according to ref 25. The proton-tofluorine distances within the crystallographic (001) plane in the apatite crystal structure are significantly longer (9.4 Å). Remarkably, that the linewidths of the components of the spectra in Figure 3b are significantly broader that those obtained by the single-pulse experiment (Figure 3a) that is explained by the selectivity of the cross-polarization based HETCOR experiment, which detects only a fraction of the 19F nuclei in the interface region characterized by a stronger disorder than the bulk phase detected in the SP experiment. To summarize the results obtained from the 19F NMR, three 19 F containing structural motifs associated with the mineral phase have been found. Besides highly crystalline fluorapatite and hydrogen-containing fluorapatite, one otheramorphouslikephase is present, in which the 19F atoms have shorter distances to protons. We assume that this phase occurs on the surface of the mineral nanoparticles in close contact to the mineral−organic interfacial region. Furthermore, other structural motifs associated with the peaks between −85 ppm and −95 ppm are found in the FAp−gelatin nanocomposite. To further elucidate the structure of fluorapatite nanocomposites, the proton local environment has been probed by 1H NMR. The single-pulse 1H MAS spectrum of FAp−gelatin nanocomposite is plotted next to the 1H HETCOR projection in Figure 4. The peak positions obtained as a result of deconvolution are listed in Table 3 and correlated to the 2D data. The 1H MAS spectrum is dominated by an asymmetric peak at 5.7 ppm, which is close to the signal at 5.8 ppm assigned to surface adsorbed water in fluorapatite.29 The broad unresolved component with low intensity between 6.5 ppm and 10 ppm results from the protons of the organic phase and acidic phosphates according to the earlier 1H studies on apatites and fluorapatites.21,27,29,30 A minor peak at 2.5 ppm is assigned to interstitial or surface adsorbed water, whose upfield shift (as compared to bulk water) points to weaker hydrogen bonding as a result of lower dimensionality of water structures. A similar peak was observed in fluorapatite and mixed hydroxyfluorapatites and attributed to highly mobile ″structural water″.29 Moreover, the 1H signals slightly upfield shifted (at 2.2 and 2.3 ppm) were detected in carbonated apatite, amorphous calcium phosphate, and deproteinated cortical bone, and assigned to isolated water molecules in the apatite channels or to channel water.31,32 Finally, the upfield peaks at 1.5 ppm and 1.1 ppm can arise from the organic phase or isolated water molecules. Due to low fractions of these components, these peaks are undetectable in the HETCOR. To further support the assignments described above, gelatin was exposed to 0.27 mol/L NaF solution. It has to be mentioned that previous investigations on oligomeric model systems showed that fluorination at position 4 of proline giving rise to 4-fluoroproline leads to enhanced conformational stability of the collagen triple helix strand and thus may lead to super stable collagen.33 The 1H SP MAS NMR spectrum of
■
DISCUSSION Based on the analysis of 31P MAS and 31P CP MAS data, at least three different 31P sites are distinguished. In the following we will interpret the 31P NMR spectra with respect to the formation of different structural arrangements. The narrow and intense line at 2.3 ppm in the SP spectrum (Figure 1a), which is only weakly associated to protons, originates from orthophosphates in the highly crystalline apatite phase (PO43− bulk). Hydroxyl groups, substituting F− within this phase and forming hydroxyfluorapatite, give rise to the corresponding 31P resonance in the 31P{1H} CP spectrum (Figure 1b). The intensities of two other peaks are strongly amplified in the 31P CP spectra pointing to closer proximity of the corresponding phosphate species to protons. The most shielded peak (δP = 1.3 ppm) results from hydrogenated phosphate groups, which can be formed either in the crystalline phase as structural defects (HPO42− bulk) or in the amorphous phase and on the mineral surface as a result of surface hydrogenation (HPO42− surface). For comparison, in our previous study on hydroxyapatite− gelatin nanocomposite11 the 31P peak at 0.9 ppm was identified as a phosphate species present at the surface of the mineral component interacting with the surrounding organic matrix. The less shielded peak (δP = 4.9 ppm) arises from deprotonated phosphate groups with remote distances to protons, but spatially associated to protons in water or in the organic phase. Therefore, we expect these structures to appear near the mineral surface and assign them to PO4 surface groups. The results of the 19F NMR measurements allow us to prove the presence of three different 19F sites: highly crystalline fluorapatite, hydroxyfluorapatite, and 19F atoms in a more disordered (amorphous) phase. However, because of the spectral overlap of the 19F signals, it is difficult to conclude the location of the first two phases with respect to each other and the corresponding domain sizes. Therefore we term it the “apatite-like phase”. It is reasonable to assume that the apatitelike phase forms the core of the mineral domains, which are covered by a more disordered (amorphous) phase, where 19F atoms are spatially associated to water molecules and acidic phosphate groups as demonstrated in the HETCOR data. The appearance of the 19F−1H correlation signals from the fluorinated organic matrix (see assignment in Table 3) proves the fact that gelatin fibrils interconnect the mineral domains forming a mineral−organic interface. Their faster build-up yields shorter distances to protons (
Intergrowth and Interfacial Structure of Biomimetic Fluorapatite− Gelatin Nanocomposite: A Solid-State NMR Study Anastasia Vyalikh,*,† Paul Simon,‡ Elena Rosseeva,‡,§ Jana Buder,‡ Rüdiger Kniep,‡ and Ulrich Scheler*,† †
Leibniz-Institut für Polymerforschung Dresden e.V., Hohe Strasse 6, 01069 Dresden, Germany Max-Planck-Institut für Chemische Physik fester Stoffe, Nöthnitzer Strasse 40, 01187 Dresden, Germany § University of Konstanz, Physical Chemistry, POB 714, D-78457 Konstanz, Germany ‡
S Supporting Information *
ABSTRACT: The model system fluorapatite−gelatin allows mimicking the formation conditions on a lower level of complexity compared to natural dental and bone tissues. Here, we report on solid-state NMR investigations to examine the structure of fluorapatite−gelatin nanocomposites on a molecular level with particular focus on organic−inorganic interactions. Using 31P, 19F, and 1H MAS NMR and heteronuclear correlations, we found the nanocomposite to consist of crystalline apatite-like regions (fluorapatite and hydroxyfluorapatite) in close contact with a more dissolved (amorphous) layer containing first motifs of the apatite crystal structure as well as the organic component. A scheme of the intergrowth region in the fluorapatite−gelatin nanocomposite, where mineral domains interact with organic matrix, is presented.
■
INTRODUCTION The biomimetic system consisting of fluorapatite (FAp) and gelatin closely resembles the biosystem hydroxyapatite− collagen, which is intimately associated with calcified hard tissues in the human body, such as bone and teeth.1,2 However, using the model system of FAp−gelatin permits us to mimic the formation conditions on a lower level of complexity than the natural formation of dental and bone tissues, and therefore contributes to a deeper understanding of the principles of biomineralization. Substitution of the rigid, insoluble collagen by water-soluble gelatin (denatured collagen) holds the chance of easy rearrangement of the protein fibers during the formation of the organic−inorganic nanocomposite. Calcium phosphate composites containing gelatin and other proteins have been extensively investigated during recent years.3−11 Fluorapatite−gelatin composite aggregates are grown by double diffusion of buffered calcium and phosphate fluoride solutions in a 300 bloom gel of gelatin. Details for their preparation are described elsewhere.12,13 Characterization using different experimental methods has shown that the resulting material is composed of hierarchically ordered spherical aggregates with fractal architecture and diameters ranging from 140 to 200 μm. The chemical composition determined by elemental analysis and FTIR spectroscopy corresponds to Ca5‑x/2(PO4)3‑x(HPO4)x(F1‑y(OH)y), 2.3 wt % gelatin, x = 0.82, 0 < y ≤ 0.1. Synchrotron X-ray diffraction on individual aggregates in different states of growth has provided information on the orientation of the mineral component.14 Local variations in the degree of mineralization across a sample have been investigated by transmission electron microscopy.15 © 2013 American Chemical Society
Particular attention has been paid to the study of pattern formation as a function of time (morphogenesis). It has been found that aggregate formation starts with elongated hexagonal prismatic seeds, followed by fractal branching and development of growing dumbbell states, and finally ends up with a slightly notched sphere.16 The “controlling factors” for the fractal development that induces the formation of dumbell-like composite structures are mainly given by the intrinsic electrical fields as proven by electron holography.7,17 So far, investigations have been extensively focused on microscopy techniques. However, a detailed study of how the inorganic phase of fluorapatite−gelatin nanocomposite interacts with the organic phase at a molecular level is lacking. The present work is a contribution to the molecular-level characterization of fluorapatite−gelatin nanocomposites in order to get molecular information on structure formation of biomimetic organic−mineral nanocomposites. Solid-state nuclear magnetic resonance (NMR) has been applied, which is sensitive to the local environment of a particular nucleus, and, therefore, is suited to study interfacial phenomena on a molecular level, whereas other structural techniques characterize the ″long-range″ order, thus giving only an average view of the structure. In this study 31P, 19F, and 1H MAS NMR and their heteronuclear combination have been performed to examine the structure of the fluorapatite−gelatin nanocomposites. Received: October 17, 2013 Revised: December 17, 2013 Published: December 19, 2013 724
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
■
Article
of 500.1 MHz for 1H, 470 MHz for 19F, and 202.5 MHz for 31P. P MAS NMR spectra were acquired with high-power proton TPPM decoupling18 using a 3.2 mm MAS probehead. Spinning frequencies of 2 kHz, 5 kHz, and 10 kHz, a single 90°-pulse of 3.2 μs duration, a recycle delay of 80 s, and 64 repetitions were applied. For the 31P{1H} cross-polarization (CP) measurements contact times of 100 to 1500 μs and a recycle delay of 5 s were used. The 19F ultrafast MAS NMR experiments were acquired using a 1.3 mm probehead with a spinning frequency of 60 kHz, a 90°-pulse duration of 5 μs, and a recycle delay of 40 s. 1 H and 19F MAS NMR spectra at 30 kHz spinning speed were measured using a 2.5 mm HFX-MAS probehead. The 19F MAS NMR experiments were acquired with a 90°-pulse duration of 5 μs and a recycle delay of 40 s. For 1H MAS NMR, the 90°-pulse duration of 3 μs, a recycle delay of 5 s, and high-power 19F decoupling using XiX composite pulse19 and rf field of 80 kHz were applied. The two-dimensional 19F−1H HETCOR experiment was performed at 30 kHz MAS using frequency-switched Lee−Goldburg (FSLG) cross-polarization with contact times of 0.1 and 1.5 ms. A recycle delay of 5 s and 1024 scans per t1 time increment were used. A total of 64 t1 slices with 22.75 μs increment in the indirect dimension were acquired. Prior to Fourier transform, apodization with 120-Hz and 30-Hz exponential line broadening was used in 19F and 1H dimensions, respectively. 1H chemical shifts were referenced to tetramethylsilane (TMS) at 0 ppm using poly(vinylidene fluoride) as an external reference; powdered ammonium dihydrogen phosphate was used to reference the 31P spectra at 0.72 ppm relative to 85 wt % H3PO4. The 19F chemical shifts were referenced relative to CFCl3 at 0 ppm using PTFE as an external reference (δF = −122 ppm). The spectra were fitted using Dmfit.20
EXPERIMENTAL SECTION Synthesis of the Samples. In a typical double diffusion experiment,6,14,15 pig skin gelatin (300 bloom, Aldrich) was heated together with water (gelatin concentration maximum 15%) under stirring for a few minutes in a water bath (after acidification with 2 N HCl). The central tube of the diffusion cell was filled with hot gelatin and left at room temperature in a vertical position until gelation was finished (for ca. 24 h). The L-shaped tubes were then attached to the central gel-tube, filled with aqueous solutions (e.g., 0.133 m CaCl2 and 0.08 m Na2HPO4/0.0027 m KF), and closed with ground glass stoppers. Both solutions were previously adjusted to the physiological pH of 7.4 with α,α,α-tris(hydroxymethyl) methylamine/HCl. The diffusion experiments were carried out between 25 and 30 °C (thermostat). Chemical Analysis. The composite aggregates were ground, washed 3 times for 20 min in distilled water at 40 °C, then centrifuged and finally dried at 40 °C in order to remove the fraction of gelatin, which is only physisorbed on the aggregates surface and not integrated in the composite. The content of calcium, phosphate, and sodium was determined by inductively coupled plasma-optical emission spectrometry (ICPOES, Varian, VISTA RL). Ten milligram samples were dissolved in 4 mL of 2 M HCl and then diluted with water in a 100 mL volumetric flask. The fluoride content was determined by using F− ion selective electrode (Mettler Toledo Inc., Wilmington, MA). The amounts of carbonate and gelatin were calculated on the basis of the C and N contents determined by use of the carrier gas hot extraction method combined with the combustion technique by means of the CHNS-analyzer 932 (LECO, USA). The sample weighting amounted to 2 mg and combustion was performed in Sn capsules under O 2 atmosphere. X-ray Diffraction. X-ray powder data were collected in transmission mode using a Huber G670 Image Plate Camera, Cu Kα1 radiation (λ = 1.540 598 Å) and germanium (111) monochromator. Lattice constants a and c of apatite were calculated by least-squares refinements using LaB6 (cubic, a = 4.156 92 Å) as internal standard. FT-IR Spectroscopy. Fourier transform infrared (FT-IR) spectra in the region of 4000−400 cm−1 were recorded at room temperature using a Bruker spectrometer (IFS 66v/S; Globar (MIR), KBr, DTGS-Detector; Program Opus/IR 3.0.3). The samples were prepared as KBr pellets (1 mg of the material under investigation dispersed in 150 mg KBr). In order to separate the overlapping bands the spectral ranges were fitted with Pseudo-Voigt 1 distributions for the bands. The software package Origin 7.0 (OriginLab Corporation, MA, USA) was used for the calculations. Raman Spectroscopy. For Raman measurements ground powder was spread on a glass plate. Raman spectra in the region of 1500−300 cm−1 were collected at room temperature using a LabRam System 010 (Horiba Jobin Yvon, France) in backscattering mode. The 632.817 nm line of a He−Ne laser with 1 mW light power was used for excitation of samples under a microscope (20-fold magnification). The spectrometer was calibrated using Si as a reference. In order to separate the overlapping bands the spectral ranges were fitted with PseudoVoigt 1 distributions for the bands. Spectra were fitted in Origin 7.0 (OriginLab Corporation, MA, USA). NMR. All NMR spectra were obtained on a (11.7 T) Bruker Avance III 500 spectrometer operating at resonance frequencies
31
■
RESULTS P MAS and 31P {1H} CP MAS NMR. Figure 1a shows the 31 P MAS NMR spectrum of FAp-gelatin nanocomposite displaying a single line centered at 2.3 ppm. Based on the literature data21−23 this signal is attributed to PO43− groups in crystalline fluorapatite. Its line width of 1.1 ppm is typical for that of pure HAp and deviates significantly from that we observed in hydroxyapatite-based nanocomposite prepared by direct precipitation in solution11 (2.3 ppm). The broadening of the latter was explained by the presence of an amorphous calcium phosphate phase located at the surfaces or at grain boundaries of the HAp nanoplatelets. To investigate the structures, which are formed at the mineral−organic interface, we applied cross-polarization (CP) NMR, which is based on the magnetization transfer from 1H to 31P sites and enables emphasizing the 31P species with strong dipole−dipole coupling to protons. The 31P{1H} CP MAS NMR spectrum measured at a spinning speed of 2 kHz to retain the information on the chemical shift anisotropy is shown in Figure 2. Fit data for three components and their assignment are summarized in Table 1. The spectrum is dominated by a peak at 2.5 ppm attributed to PO43− groups in crystalline fluorapatite. Its small chemical shift anisotropy (CSA) of about 16 ppm is a result of a high symmetry in an isolated PO4 tetrahedron reflected by four nearly identical P−O bond lengths. Two weaker shoulder peaks at 1.3 ppm and 4.9 ppm have been assigned to protonated HPO42− and unprotonated PO43− phosphate surface groups, respectively, according to the previous assignment in synthetic 725
31
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
calcium phosphates, supported by the difference in their CSA values.21,24 The smaller CSA value of −17 ppm for unprotonated PO43− (surface) reflects the most likely tetrahedral phosphorus coordination. This orthophosphate group can be located at the mineral surface. In contrast, the larger CSA for the protonated HPO4 sites (62 ppm) points to the distortion of symmetry compared to PO43− tetrahedra yielding either HPO4 (surface) or HPO42− (bulk) groups, when the latter may replace the PO43− (bulk) in the apatite crystal structure. The CP build-up curves for each spectral component resolved in the CP spectra (Figure 1c) confirm the proposed assignment demonstrating the slower build-up for the peak at 2.5 ppm characteristic of apatite PO43− groups in contrast to other two signals, which reach the maximum at ca. 1 ms showing closer proximity to the protons. Complementary studies have proven the existence of HPO4 groups, which appeared as small peaks and shoulders around 874 cm−1 and 1006 cm−1 in the Raman spectrum (see SI Figure S1) and as peaks around 545 cm−1 and 864 cm−1 in the IR spectrum (see SI Figure S2). Chemical analysis of the FAp−gelatin nanocomposite has revealed the amount of hydrogen phosphate in the bulk phase to be about 10 mol %, which is not in agreement with 62% found from NMR (see Table 1). NMR indeed reflects the amount of phosphorus-containing species in the near-surface or interface region, but not the composition of the entire sample. This becomes visible if one compares the centerband 31P CP NMR spectrum at 10 kHz (Figure 1b) with the single pulse (SP) spectrum (Figure 1a). The asymmetric lineshape and significant broadening of the 31P CP spectrum has been explained by the selectivity of the CP experiment, which detects only a fraction of the 31P nuclei, which is in close proximity to protons, and whose fraction is significantly smaller with respect to all phosphorus atoms in the bulk detected in the 31 P SP experiment. Therefore, upon proper scaling of intensity the SP spectra shall completely conceal the tiny contribution from 31P atoms, which are observed in the CP spectra. 19 F and 1H MAS NMR. The 19F MAS spectrum at 60 kHz is shown in Figure 3a. Upon deconvolution, four components are discerned, whose fit parameters are summarized in Table 2. The peak at −102.8 ppm is known to arise from crystalline fluorapatite.23 The second component at −103.8 ppm can be assigned to apatite-like structures containing hydroxyl groups or isolated water molecules in apatite channels, based on the correlation of 19F chemical shift parameters and the F−/OH− ratio in mixed hydroxyfluorapatite reported in ref 25. The similarity of the shift tensor parameters of two peaks discussed above points to their similarities in the crystal structure and the local 19F environment. In contrast, the strong increase of the 19 F line width of the latter (4.9 ppm vs 1.6 ppm in pure crystalline fluorapatite) results from an inhomogeneous broadening due to structural disorder caused by the presence of the proton-containing species. The broad peak at −108.5 ppm is typical for Ca···F interactions, for example, in CaF2,26 although
Figure 1. 31P MAS (10 kHz) NMR spectra of FAp−gelatin nanocomposite detected using (a) direct polarization and (b) crosspolarization from 1H together with the deconvolution results. Spectra are normalized to a maximum intensity. The CP spectrum is fitted with three spectral components with the same isotropic shift parameters as used in the slow MAS spectrum (Table 1, Figure 2), but with the varied intensities and linewidths caused by folding back the spinning side bands; (c) CP build-up for three components indicated in Table 1 and (b) (squares denote the signal at 1.3 ppm, circles at 2.5 ppm, triangles at ca. 5 ppm).
Figure 2. Broad width 31P{1H} CP MAS NMR spectrum at a spinning speed 2 kHz and 1 ms contact time of FAp−gelatin nanocomposite (top) and its deconvolution (bottom). The green line denotes the component assigned to apatitic PO43−; the red and blue ones arise from unprotonated and protonated surface phosphate groups, respectively. The difference between the generated fit and the experimental spectrum is shown underneath.
Table 1. Fit Data of 31P{1H} CP MAS NMR Spectruma chemical shift tensor, ppm
a
δiso
δani
η
width, ppm
I, %
assignment
region
1.3 ± 0.3 2.5 ± 0.1 4.9 ± 0.3
62 ± 5 16 ± 3 −17 ± 3
0.4 ± 0.2 0.5 ± 0.2 0.2 ± 0.1
4.1 ± 0.2 2.3 ± 0.2 2.6 ± 0.5
62 ± 5 27 ± 5 11 ± 5
HPO42− PO43− PO43−
bulk/surface bulk surface
MAS 2 kHz, contact time 1 ms. 726
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
Figure 4. 19F−1H HETCOR spectrum of FAp−gelatin nanocomposite. The spectra on the top are 19F MAS and the horizontal projection of the 2D spectrum. On the right are the single-pulse 1H MAS (30 kHz) spectrum and the 1H projection of the 2D spectrum.
Table 3. 1H Isotropic Chemical Shifts δH (±0.1 ppm) as Measured from 1H−19F HETCOR with Correlation to the 19 F Chemical Shifts (δH, ppm /δF, ppm) and 1H MAS NMR
Figure 3. (a) 19F MAS (60 kHz) NMR spectrum of FAp-gelatin nanocomposite. (b) Projection (total signal intensity) of the 19F−1H HETCOR spectra in 19F direction at contact time of 0.1 and 0.5 ms. Colored lines represent decomposition and the gray line denotes the difference between the generated fit and the experimental spectra.
1 H−19F HETCOR
2.9/−103.1 6.6/−103.6 6.6/−110.5 7.1/−85.5 7.1/−95.5 11.1/−99 ca. 13/−110
Table 2. Fit Parameters of 19F MAS NMR Spectrum at 60 kHz chemical shift tensor, ppm δiso (±0.1)
δani (±5 ppm)
−95.7 −102.8 −103.8 −108.6
35 52 53 35
η
width, ppm (±0.5)
intensity, % (±5)
assignment
0.5 0.7 0.7 0.5
9.2 1.6 4.9 5.1
6 18 49 27
organic phase fluorapatite hydroxyfluorapatite amorphous phase
1
H MAS NMR
Assignment
1.1 and 1.5 2.5 5.7 6.1 6.5−10.0 -
organic phase or isolated water isolated water close to apatite-like phase water bound to organic phase water bound to organic phase fluorinated organic phase fluorinated organic phase strongly H-bonded protons HPO4 (surface/bulk) groups
containing (OH, HPO4) fluorapatite structures. In contrast, the peak at −102.8 ppm, which is observed in the 19F SP MAS NMR spectrum (Figure 3a), does not appear here, confirming its assignment to pure crystalline FAp. In the 2D spectrum, it is easy to see that both 19F signals at −103.6 ppm and −110.5 ppm correlate with the 1H peak at 6.6 ppm. The chemical shift of the 1H peak is too high to be attributed to bulk or surface water, but could be related to water molecules strongly hydrogen bonded to, most probably, protein molecules. Additionally, two other overlapping signals, however, with well-defined maxima arise at 7.1 ppm/−85.5 ppm and 7.1 ppm/−95.5 ppm. For their assignment, we refer to literature data, where close 1H signals in relevant systems have been observed. Thus, in 1H CRAMPS spectra of bovine collagen and animal bone27 as well as in a 1H−31P HETCOR spectrum of synthetic hydroxyapatite−gelatin nanocomposite reported in our previous work,11 the peaks at ca. 7 ppm have been assigned to the organic phase. Therefore, the correlation peaks at −85.5 ppm and −95.5 ppm detected in the present work can arise either from the fluorinated gelatin or from new structures on the mineral surface caused by the presence of the organic phase. Finally, the minor peak observed at 2.9 ppm/−103.1 ppm is assigned to isolated water molecules in the crystalline FAp, i.e., apatite-like channel water. It should be mentioned that similar 1 H signals at 3.2 ppm and 6.6 ppm have also been reported for low−OH− fluor−chlorapatite28 and were attributed to 1H atoms in a separate phase, not associated with apatite. Comparison of the HETCOR projections in the 19F direction at two different contact times (Figure 3b) yields the information about 1H− 19 F dipole interactions and an
no indication for the presence of CaF2 has been found in the corresponding Raman spectrum. Additionally, X-ray powder diffraction data are also consistent with the absence of CaF2 (see SI Figure S3). The line broadening of the signal at −108.5 ppm may originate from structural disorder resulting in a broad distribution of isotropic chemical shifts, as well as from strong homo- and heteronuclear dipolar coupling. Since application of high power 1H decoupling and the variation of the MAS speed (see SI Figure S4), expected to evidence the presence of the dipolar interactions, did not exhibit an effect on the linewidth, we concluded that structural disorder is the main origin of the inhomogeneous line broadening. To investigate the nature of the peaks at −108.5 ppm and −95.7 ppm, a further twodimensional 19F−1H heteronuclear correlation (HETCOR) experiment has been performed, which provides information on the spatial correlation of 1H containing groups in the inorganic phase and organic matrix to the 19F nuclei initially assumed to be present exclusively in the mineral phase. The two-dimensional 19 F− 1 H HETCOR spectrum (Figure 4) shows one asymmetric broad cross signal with the maximum intensity at δH/δF = 6.6 ppm/−110.5 ppm. For accurate analysis of the 2D spectrum we inspected individual slices in both directions. The chemical shifts of the correlation peaks are summarized in Table 3. The second dominating contribution is visible at 6.6 ppm/−103.6 ppm. We have to emphasize that the appearance of the peak at −103.6 ppm in the 2D experiment confirms its assignment to hydrogen 727
dx.doi.org/10.1021/jp410299x | J. Phys. Chem. B 2014, 118, 724−730
The Journal of Physical Chemistry B
Article
gelatin exposed to fluoride solution (see SI Figure S5a) is similar to that of protonated gelatin,34 whereas the 1H{19F} CP NMR experiment (see SI Figure S5b) enables resolving the 1H peaks at 7.2 and 11.6 ppm, which are very close to those observed in the HETCOR data (7.1 and 11.1 ppm). This observation corroborates their assignment to the organic phase. Fluorination of gelatin fibers has been proven in the 19F spectra (see SI Figure S5c,d). Although a very poor signal-to-noise ratio resulting from a low degree of fluorination hinders unequivocal assignment of 19F resonances in gelatin, several 19F signals can be distinguished. Comparing the SP and CP 19F spectra (Figure S5c,d), one can see that the signal at ca. −98 ppm shows the closest proximity to protons and thus can be attributed to the organic phase. This result confirms our assignment of the HETCOR data presented in Table 3 and supports a scenario of partial fluorination of gelatin during the formation of the nanocomposite.
estimation of associated protons-to-fluorine proximities, and therefore provides a cross-check toward the assignment of 19F signals. Apparently, the signals at ca. −85 ppm, −95 ppm, and −110 ppm have faster cross-polarization build-up translating the stronger 1H−19F dipole interactions than that in the apatitelike phase (at ca. −103 ppm). Thus, estimation of the build-up of the former signals points to 1H···19F distances shorter than 2 Å. It is worth to note that the distances of spatially separated F− and OH− ions in substituted fluorapatite are 2.03 Å for a F··· F···H group and 2.12 Å to 2.18 Å for a H···F···H groups along the crystallographic c axis according to ref 25. The proton-tofluorine distances within the crystallographic (001) plane in the apatite crystal structure are significantly longer (9.4 Å). Remarkably, that the linewidths of the components of the spectra in Figure 3b are significantly broader that those obtained by the single-pulse experiment (Figure 3a) that is explained by the selectivity of the cross-polarization based HETCOR experiment, which detects only a fraction of the 19F nuclei in the interface region characterized by a stronger disorder than the bulk phase detected in the SP experiment. To summarize the results obtained from the 19F NMR, three 19 F containing structural motifs associated with the mineral phase have been found. Besides highly crystalline fluorapatite and hydrogen-containing fluorapatite, one otheramorphouslikephase is present, in which the 19F atoms have shorter distances to protons. We assume that this phase occurs on the surface of the mineral nanoparticles in close contact to the mineral−organic interfacial region. Furthermore, other structural motifs associated with the peaks between −85 ppm and −95 ppm are found in the FAp−gelatin nanocomposite. To further elucidate the structure of fluorapatite nanocomposites, the proton local environment has been probed by 1H NMR. The single-pulse 1H MAS spectrum of FAp−gelatin nanocomposite is plotted next to the 1H HETCOR projection in Figure 4. The peak positions obtained as a result of deconvolution are listed in Table 3 and correlated to the 2D data. The 1H MAS spectrum is dominated by an asymmetric peak at 5.7 ppm, which is close to the signal at 5.8 ppm assigned to surface adsorbed water in fluorapatite.29 The broad unresolved component with low intensity between 6.5 ppm and 10 ppm results from the protons of the organic phase and acidic phosphates according to the earlier 1H studies on apatites and fluorapatites.21,27,29,30 A minor peak at 2.5 ppm is assigned to interstitial or surface adsorbed water, whose upfield shift (as compared to bulk water) points to weaker hydrogen bonding as a result of lower dimensionality of water structures. A similar peak was observed in fluorapatite and mixed hydroxyfluorapatites and attributed to highly mobile ″structural water″.29 Moreover, the 1H signals slightly upfield shifted (at 2.2 and 2.3 ppm) were detected in carbonated apatite, amorphous calcium phosphate, and deproteinated cortical bone, and assigned to isolated water molecules in the apatite channels or to channel water.31,32 Finally, the upfield peaks at 1.5 ppm and 1.1 ppm can arise from the organic phase or isolated water molecules. Due to low fractions of these components, these peaks are undetectable in the HETCOR. To further support the assignments described above, gelatin was exposed to 0.27 mol/L NaF solution. It has to be mentioned that previous investigations on oligomeric model systems showed that fluorination at position 4 of proline giving rise to 4-fluoroproline leads to enhanced conformational stability of the collagen triple helix strand and thus may lead to super stable collagen.33 The 1H SP MAS NMR spectrum of
■
DISCUSSION Based on the analysis of 31P MAS and 31P CP MAS data, at least three different 31P sites are distinguished. In the following we will interpret the 31P NMR spectra with respect to the formation of different structural arrangements. The narrow and intense line at 2.3 ppm in the SP spectrum (Figure 1a), which is only weakly associated to protons, originates from orthophosphates in the highly crystalline apatite phase (PO43− bulk). Hydroxyl groups, substituting F− within this phase and forming hydroxyfluorapatite, give rise to the corresponding 31P resonance in the 31P{1H} CP spectrum (Figure 1b). The intensities of two other peaks are strongly amplified in the 31P CP spectra pointing to closer proximity of the corresponding phosphate species to protons. The most shielded peak (δP = 1.3 ppm) results from hydrogenated phosphate groups, which can be formed either in the crystalline phase as structural defects (HPO42− bulk) or in the amorphous phase and on the mineral surface as a result of surface hydrogenation (HPO42− surface). For comparison, in our previous study on hydroxyapatite− gelatin nanocomposite11 the 31P peak at 0.9 ppm was identified as a phosphate species present at the surface of the mineral component interacting with the surrounding organic matrix. The less shielded peak (δP = 4.9 ppm) arises from deprotonated phosphate groups with remote distances to protons, but spatially associated to protons in water or in the organic phase. Therefore, we expect these structures to appear near the mineral surface and assign them to PO4 surface groups. The results of the 19F NMR measurements allow us to prove the presence of three different 19F sites: highly crystalline fluorapatite, hydroxyfluorapatite, and 19F atoms in a more disordered (amorphous) phase. However, because of the spectral overlap of the 19F signals, it is difficult to conclude the location of the first two phases with respect to each other and the corresponding domain sizes. Therefore we term it the “apatite-like phase”. It is reasonable to assume that the apatitelike phase forms the core of the mineral domains, which are covered by a more disordered (amorphous) phase, where 19F atoms are spatially associated to water molecules and acidic phosphate groups as demonstrated in the HETCOR data. The appearance of the 19F−1H correlation signals from the fluorinated organic matrix (see assignment in Table 3) proves the fact that gelatin fibrils interconnect the mineral domains forming a mineral−organic interface. Their faster build-up yields shorter distances to protons (
